You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thank you for this feature suggestion, @Liripo! I appreciate your input on exposing the base_url parameter for each model.
I'll definitely consider implementing this enhancement in an upcoming release. This would provide more flexibility for users, especially those who need to connect to alternative endpoints or proxy services.
In the meantime, I'd like to highlight a few current options that might help address your needs:
For users in mainland China, we already support several providers that work well within China, including:
DeepSeek
Qwen (Alibaba)
MiniMax
Stepfun
Zhipu
We've also recently added OpenRouter support, which provides a unified API to access multiple models through a single endpoint. This can be a convenient solution for accessing various models without managing multiple API endpoints.
Additionally, we just added Azure OpenAI support, which allows for custom endpoints through the Azure platform.
Using the ellmer.tidyverse.org service is an interesting suggestion that I'll explore further. It could potentially simplify access for some users.
I've added this to our feature roadmap and will update this issue when we make progress on implementation. If you have any specific use cases or additional requirements for this feature, please feel free to share them here.
Thanks again for contributing to the improvement of mLLMCelltype!
每个模型的 base_url 如果能够暴露就好了。
也许可以直接调用 https://ellmer.tidyverse.org/ 会更方便点。
The text was updated successfully, but these errors were encountered: